I N S T I T U T E
Deterring Dictatorship: Explaining Democratic Resilience since 1900
Vanessa A. Boese, Amanda B. Edgell, Sebastian Hellmeier, Seraphine F. Maerz, Sta�an I. Lindberg
Working Paper SERIES 2020:101
THE VARIETIES OF DEMOCRACY INSTITUTE
May 2020
Varieties of Democracy (V-Dem) is a new approach to conceptualization and measurement of democracy. The headquarters – the V-Dem Institute – is based at the University of Gothenburg with 19 staff. The project includes a worldwide team with six Principal Investigators, 14 Project Managers, 30 Regional Managers, 170 Country Coordinators, Research Assistants, and 3,000 Country Experts. The V-Dem project is one of the largest ever social science research-oriented data collection programs.
Please address comments and/or queries for information to:
V-Dem Institute
Department of Political Science
University of Gothenburg
Sprängkullsgatan 19, PO Box 711
SE 40530 Gothenburg
Sweden
E-mail: [email protected]
V-Dem Working Papers are available in electronic format at www.v-dem.net.
Copyright ©2020 by authors. All rights reserved.
Deterring Dictatorship:
Explaining Democratic Resilience since 1900*
Vanessa A. Boese1, Amanda B. Edgell1, Sebastian Hellmeier1,
Seraphine F. Maerz1, and Staffan I. Lindberg1
1V-Dem Institute, Department of Political Science, University of Gothenburg
* Corresponding author: Vanessa A. Boese ([email protected]). Funding: We recognize support by the Swedish Research Council, Grant 2018-01614, PI: Anna Lührmann; by Knut and Alice Wallenberg Foundation to Wallenberg Academy Fellow Staffan I. Lindberg, Grant 2018.0144; by European Research Council, Grant 724191, PI: Staffan I. Lindberg; as well as by internal grants from the Vice- Chancellor’s office, the Dean of the College of Social Sciences, and the Department of Political Science at University of Gothenburg. The computations of expert data were enabled by the Swedish National Infrastructure for Computing (SNIC) at National Supercomputer Centre, Linköping University, partially funded by the Swedish Research Council through grant agreement no. 2019/3-516.
Abstract
Democracy is under threat globally from democratically elected leaders engaging in erosion
of media freedom, civil society, and the rule of law. What distinguishes democracies that
prevail against the forces of autocratization? This article breaks new ground by
conceptualizing democratic resilience as a two-stage process, whereby democracies first
exhibit resilience by avoiding autocratization altogether and second, by avoiding democratic
breakdown given that autocratization has occurred. To model this two-stage process, we
introduce the Episodes of Regime Transformation (ERT) dataset tracking autocratization
since 1900. These data demonstrate the extraordinary nature of the current wave of
autocratization: Fifty-nine (61%) episodes of democratic regression in the ERT began after
1992. Since then, autocratization episodes have killed an unprecedented 36 democratic
regimes. Using a selection-model, we simultaneously test for factors that make democracies
more prone to experience democratic regression and, given this, factors that explain
democratic breakdown. Results from the explanatory analysis suggest that constraints on the
executive are positively associated with a reduced risk of autocratization. Once
autocratization is ongoing, we find that a long history of democratic institutions, durable
judicial constraints on the executive, and more democratic neighbours are factors that make
democracy more likely to prevail.
3
Introduction
Democracy is under threat globally. Over 20% of the world’s polities and one-third of the
global population are now experiencing substantial and sustained declines in democracy.2 As
this article demonstrates, democracies are particularly vulnerable during this “third wave” of
autocratization.3 Of the 96 episodes of autocratization within democracies between 1900 and
2019, 59 (61%) began after 1992. Since then, autocratization has killed an unprecedented 36
democratic regimes, causing over 700 million people to lose access to democratic institutions.
Rather than employing blatant and unconstitutional means (e.g. military or self-coups),4
democratically elected leaders increasingly engage in more subtle and nuanced attacks on
democratic institutions and practices, such as executive aggrandizement,5 curtailment of
media freedoms,6 and the gradual erosion of horizontal accountability.7 Freedom of
expression and civil society are typically affected first and the most.8
What distinguishes democracies that prevail against a global wave of autocratization
from those that do not? Our object of inquiry is democratic resilience – the capacity to
prevent substantial regression in the quality of democratic institutions and practices.9
Democratic resilience takes two forms. Democracies can prevent autocratization altogether,
meaning they never experience a substantial or sustained decline in democratic qualities (such
as New Zealand and Sweden). Alternatively, democracies may experience an episode of
autocratization but pro-democracy actors and institutions manage to change the course and
avert democratic breakdown (such as South Korea from 2008–2016, and Benin from 2007–
2012).
These two forms of democratic resilience are conceptually and empirically distinct.
For democracies, episodes of autocratization are uncommon. We find only 96 such episodes
in 70 polities from 1900 to 2019. This suggests that a given democratic country-year exhibits
an overall high resilience, all else equal. However, once a democracy enters an autocratization
episode, democratic resilience becomes unlikely. A mere 19 episodes of democratic regression
managed to avert breakdown. Thus, democratic resilience manifests as a two-stage
2 Maerz et al., “State of the world 2019: Autocratization Surges - Resistance Grows.” 3 Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?” 4 Svolik, “Which democracies will last? Coups, incumbent takeovers, and the dynamic of democratic consolidation.” 5 Bermeo, “On democratic backsliding.” 6 Mechkova, Lührmann, and Lindberg, “How much democratic backsliding?” 7 Coppedge, “Eroding regimes: What, where, and when?” 8 Maerz et al., “State of the world 2019: Autocratization Surges - Resistance Grows.” 9 This builds on concepts of democratic durability and resilience as defined by Burnell and Calvert, “The resilience of democracy: An introduction.”
4
phenomenon, where some democracies show resilience by avoiding autocratization
altogether, and others, after having “selected into” an episode of autocratization, prevail by
averting breakdown.
Existing studies typically address this two-stage process by modelling democratic
breakdown as a discrete outcome, ignoring potentially important selection effects in the
process.10 Alternatively, building on Lührmann and Lindberg,11 we take an episodes approach
that treats autocratization as a process. To do so, we use the V-Dem data12 to develop the
Episodes of Regime Transformation (ERT) dataset, covering democratization and
autocratization episodes in most political units from 1900 to 2019.13 The episode approach
enables us to empirically model the two-stage process of democratic resilience using a
Heckman selection model. In the first “selection” stage model, we assess which factors are
associated with resilience to experiencing an autocratization episode. In the second
“outcome” stage, we analyse what factors are associated with resilience to democratic
breakdown, conditional on being in an episode of democratic regression.
We offer several novel insights into democratic resilience. First, our data provides the
most comprehensive coverage of autocratization episodes from 1900 to 2019, and for the
first time, codes subtypes of democratic regression and autocratic regression and outcomes
of these episodes. This approach generates new information on the scope and nature of
autocratization. We find that democracies are increasingly susceptible to undergoing
autocratization and that once autocratization begins, avoiding democratic breakdown is very
rare. Second, our empirical models suggest that democracies with stronger constraints on the
executive are both less likely to undergo democratic regression and if they do, they are more
likely to avert breakdown. High levels of economic development in democracies also reduce
the probability of experiencing an episode of autocratization. In addition, we find that
neighbourhood levels of democracy play a crucial role; the breakdown of democracy is less
likely in regions where democratic institutions are the norm. Finally, countries with a long
democratic experience are also more resilient to breakdown once democratic regression is
ongoing.
10 E.g. Svolik, “Authoritarian reversals and democratic consolidation”; Bernhard, Nordstrom, and Reenock, “Economic performance, institutional intermediation, and democratic survival”; Alemán and Yang, “A duration analysis of democratic transitions and authoritarian backslides” 11 Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?” 12 Coppedge et al., “V-Dem Dataset V10” 13 Combined with our coding of episodes of democratization, their subtypes, and outcomes, the ERT dataset provides a useful new tool for scholars to explore regime transformation since 1900. Maerz et al., “Vdemdata – and R package to load, explore and work with the most recent V-Dem (Varieties of Democracy) dataset”; Edgell et al. “Episodes of Regime Transformation (ERT) dataset codebook, v1.0”.
5
Existing insights into democratic resilience
The literature on democratic breakdown and survival informs much of what we know about
democratic resilience. Scholars in this field typically test for the effects of structural
determinants on the probability of democratic survival or breakdown as events.14
Alternatively, they may look at more incremental regressions (sometimes termed backsliding,
reversal, or erosion) using annual changes in measures of democracy.15 We focus here on four
main determinants from the literature: institutional constraints on the executive, economic
factors, neighbouring regimes, and previous democratic experience.
Constraints on the executive
A prominent body of work concerns the “perils of presidentialism”.16 According to Linz,
separate legislative and executive elections create a dual legitimacy and individual mandate of
the executive that predisposes political actors to view presidential systems as a zero-sum
game. This discourages coalitions and a diversity of viewpoints, while concentrating
substantial powers in one individual.17 In effect, presidential systems are more prone to
political polarization and deadlock, personalization of politics, and exclusion of losers when
the winner takes-it-all, thus furthering military coups and other types of breakdown compared
to parliamentary democracies.18
Noting that the United States is the only durable presidential democracy,19 several
large-N studies find a negative relationship between presidentialism and rates of democratic
survival.20 Case evidence suggests that executives in presidential democracies are likely to
“rule at the edge of the constitution” because the legislature has limited removal powers.21 In
14.E.g. Svolik, “Authoritarian reversals and democratic consolidation”; Hollyer, Rosendorff and Vreeland, “Transparency, Protest and Democratic Stability”; Bernhard, Nordstrom, and Reenock, “Economic performance, institutional intermediation, and democratic survival” 15 E.g. Ginsburg and Huq, “How to save a constitutional democracy”; Erdmann, “Transition from Democracy. Loss of Quality, Hybridisation and Breakdown of Democracy”; Mechkova, Lührmann, and Lindberg, “How much democratic backsliding?” 16 Linz, “The perils of presidentialism.” 17 Linz, “The breakdown of democratic regimes”; Linz, “The perils of presidentialism”; Linz and Valenzuela, “The failure of presidential democracy” 18 Kaufman and Haggard, “Democratic decline in the United States: What can we learn from middle-income backsliding?” 19 E.g. Linz, “The perils of presidentialism”; Cheibub, Presidentialism, parliamentarism, and democracy. 20 E.g. Mainwaring, “Presidentialism, multipartism, and democracy: The difficult combination”; Bernhard, Nordstrom, and Reenock, “Economic performance, institutional intermediation, and democratic survival”; Riggs, “Presidentialism versus parliamentarism: Implications for representativeness and legitimacy”; Svolik, “Which democracies will last? Coups, incumbent takeovers, and the dynamic of democratic consolidation.” 21 Stepan and Skach, “Constitutional frameworks and democratic consolidation: Parliamentarianism versus presidentialism.”
6
response, scholars and practitioners often promote parliamentarism to strengthen democratic
endurance, particularly in developing countries and divided societies.22 However, several
critiques contend that the relationship is spurious, driven by presidentialism being adopted in
countries already susceptible to democratic breakdown due to prior experiences with military
rule23 or democratic instability.24 Other scholars warn that an uncritical embrace of
parliamentarism in ethnically-divided countries may not produce the desired effect.25
Recent trends suggest that attacks on democracy are often driven by a concentration
of power in democratically elected executives. This calls for revisiting Linz’s opening
statement, which centres on the effects of weak constraints on the executive. Empirically, the
extent to which the executive is constrained de facto varies considerably both within and
between systems of government. Executive aggrandizement affects both presidential and
parliamentary systems, as the examples of Hungary and Poland currently illustrate.26 In effect,
the phenomenon of “presidential hegemony” poses a potential risk to democratic stability
everywhere.27
The Linz-thesis is yet to be tested using granular data on the specific causal
mechanism of weak constraints on the executive.26 Rather than relying solely on de jure
institutions, the V-Dem data provides the opportunity to test this theory using measures of
de facto powers. Our expectation is that stronger constraints on the executive by the legislature
and the judiciary are positively associated with both a lower likelihood of autocratization
episodes in democracies and greater resilience to democratic breakdown once such an episode
has begun.
Economic factors
Since Lipset’s seminal work on the societal effects of economic development, questions about
the links between economics and democratic stability have preoccupied the discipline.28 While
many view Lipset as the birthplace of the modernization theory, his original focus is actually
on democratic resilience, arguing that “the more the well-to-do a nation, the greater the
22 Lijphart, Democracy in plural societies: A comparative exploration; Lijphart, “Constitutional design for divided societies.” 23. Cheibub, Presidentialism, parliamentarism, and democracy. 24 Hiroi and Omori, “Perils of parliamentarism? Political systems and the stability of democracy revisited.” 25 E.g. Wilson, “A Closer Look at the Limits of Consociationalism.” 26 Bermeo, “On democratic backsliding.” 27 Pérez-Liñán, Schmidt, and Vairo, “Presidential hegemony and democratic backsliding in Latin America, 1925–2016.” 28 Lipset, “Some social requisites of democracy: Economic development and political legitimacy.”
7
chances that it will sustain democracy” (emphasis added).29 Some tests of Lipset’s theory such
as by Przeworski and Limongi suggest that democracies become resilient to breakdown once
they are above a certain threshold level of income - at the time $6,000 (GNP/cap, PPP).30
Several studies find that positive economic growth predicts democratic survival,31 but this
may be good for the stability of any regime, including autocracies.32 Overall, the expectation
is that a better quality of life makes people more likely to support the status quo over those
seeking to undo the existing order.
Indicators of economic development are now standard practice in models estimating
democratization, democratic breakdown, and democratic survival.33 In line with the bulk of
previous studies, we expect that higher levels of economic development will make
democracies more resilient to experiencing an autocratization episode. Once a democracy
selects into an episode of autocratization, we remain agnostic about the potential stabilizing
effects of development.
Neighbourhood effects
Several studies provide evidence of diffusion effects across countries. This is often described
as a “pull towards the regional mean” – or a tendency for countries “left behind” to eventually
adapt to regional norms about institutional configurations for autocratic as well as democratic
regimes by way of diffusion, emulation, spill-over, or demonstration effects.34 In light of the
29 Ibid., 75. 30 For example, Burkhart and Lewis-Beck, “Comparative Democracy: The Economic Development Thesis”; Przeworski et al., Democracy and development: Political institutions and well-being in the world, 1950-1990; Boix and Stokes, “Endogenous democratization”; Epstein et al., “Democratic transitions”; Teorell, Determinants of democratization: Explaining regime change in the world, 1972–2006. Przeworski and Limongi, “Modernization: Theories and facts,” 165. 31 For instance, Gasiorowski, “Economic Crisis and Political Regime Change: An Event History Analysis”; Gates et al., “Institutional Inconsistency and Political Instability: Polity Duration, 1800– 2000”; Morlino and Quaranta, “What is the impact of the economic crisis on democracy? Evidence from Europe.” 32 E.g. Alemán and Yang, “A duration analysis of democratic transitions and authoritarian backslides”; Svolik, “Authoritarian reversals and democratic consolidation”; Feng, “Democracy, political stability and economic growth”; Gates et al., “Institutional Inconsistency and Political Instability: Polity Duration, 1800–2000.” 33 For example Olson, “Dictatorship, Democracy, and Development”; Teorell, Determinants of democratization: Explaining regime change in the world, 1972–2006; Morlino and Quaranta, “What is the impact of the economic crisis on democracy? Evidence from Europe.” 34 E.g. Brinks and Coppedge, “Diffusion is no illusion: Neighbor emulation in the third wave of democracy”; Gleditsch and Ward, “Diffusion and the international context of democratization”; Tansey, Koehler, and Schmotz, “Ties to the Rest: Autocratic Linkages and Regime Survival”; Bader, Grävingholt, and Kästner, “Would autocracies promote autocracy? A political economy perspective on regime-type export in regional neighbourhoods”; Risse and Babayan, “Democracy promotion and the challenges of illiberal regional powers: introduction to the special issue”; Gleditsch, All international politics is local: The diffusion of conflict, integration, and democratization; Gleditsch and Ward, “Diffusion and the international context of democratization”; Houle and Kayser, “The Two-step Model of Clustered Democratization.”; Gates et al., “Institutional Inconsistency and Political Instability: Polity Duration, 1800–2000.”
8
gradual nature of democratic regression during the third wave, we expect at most small
neighbourhood effects on the probability of experiencing democratic regression. Once a
democracy opts into an episode of autocratization, however, we hypothesize that the
complete dismantling of democracy becomes more difficult for aspiring autocrats in more
democratic regions.
Previous democratic experience
Previous experience under democracy may reinforce democratic survival through the
“construction of solid links between the democratic institutions and society”.35 Some scholars
suggest that the institutionalization of democratic procedures like party systems and judicial
institutions36 helps to address “problems of monitoring and social coordination that
complicate democratic compromise”.37 Others claim that election cycles have a self-
reinforcing, self-improving quality, altering the incentives to accept the rules of the game.38
Indeed, everyday experiences living under democracy seem to promote democratic attitudes
within society, making successful challenges to democracy less likely over time.39 Based on
these findings, we expect that previous experience with democracy will decrease the
likelihood of democratic regression. If a democracy nevertheless experiences autocratization,
we hypothesize that countries with longer democratic histories will show a greater resilience
to democratic breakdown.
An episodes approach to democratic resilience
While some partial exceptions exist,40 previous insights on democratic resilience tend to rely
on a conceptualization of regime transitions as events, i.e. democratic breakdowns occur at a
specific point in time and democratic survival or durability is indicated by the absence of a
breakdown in a given year. This approach overlooks the important conceptual distinction
between two forms of democratic resilience - the complete avoidance of autocratization
35 Tomini and Wagemann, “Varieties of contemporary democratic breakdown and regression: A comparative analysis,” 690. 36 Pérez-Liñán and Mainwaring, “Regime legacies and levels of democracy: evidence from Latin America.” 37 Reenock, Staton, and Radean, “Legal institutions and democratic survival,” 491. 38 Lindberg, Democracy and elections in Africa, 144. 39 Putnam, Making Democracy Work: Civic Traditions in Modern Italy; Persson and Tabellini, “Democratic capital: the nexus of political and economic change”; Guiso, Sapienza, and Zingales, “Long-term persistence”; Grosfeld and Zhuravskaya, “Cultural vs. economic legacies of empires: Evidence from the partition of Poland.” 40 Erdmann, “Transition from Democracy. Loss of Quality, Hybridisation and Breakdown of Democracy”; Ginsburg and Huq, “How to save a constitutional democracy.”
9
altogether and the ability to avert breakdown once autocratization has begun. A rich
comparative literature suggests that democratic breakdowns are the culmination of episodes
of autocratization – a process of regime transformation producing substantial declines in
democracy over a sustained period. In democracies, autocratization entails intra- and inter-
elite bargaining between regime insiders and opposition forces.41 This can unfold over an
extended period and does not always end with a complete transition to autocracy.42 Focusing
on democratic breakdowns as events could “blind us to potentially important and
theoretically revealing cases.”43 This is relevant today, as the world experiences a wave of
autocratization characterized by gradual regressive reforms in many democracies whose fates
as of yet remain uncertain.44 For these reasons, we adopt the episodes approach to studying
regime transformation.
Conceptualizing episodes of autocratization
Based on Lührmann and Lindberg, we define autocratization as any “substantial de-facto
decline of core institutional requirements for electoral democracy.”45 This definition is
encompassing, allowing for autocratization to occur in both democratic and autocratic
regimes.46 We speak of episodes of autocratization to capture periods with a definitive start and
end date during which substantial and sustained declines in the quality of democracy take
place.47 These transformations may be incremental and may not necessarily yield a complete
transition between democracy and autocracy. This allows us to capture the full range of
possibilities when it comes to democratic resilience, combining the benefits of studying
autocratization as a process unfolding over time with those achieved through “breakdown as
event”-models.
41 E.g. Linz, The breakdown of democratic regimes; Bermeo, Ordinary people in extraordinary times: The citizenry and the breakdown of democracy; Mainwaring and Pérez-Liñán, Democracies and dictatorships in Latin America: emergence, survival, and fall. 42 E.g. Tilly, Contention and democracy in Europe, 1650-2000; Epstein et al., “Democratic transitions”; Linz and Stepan, Problems of democratic transition and consolidation: Southern Europe, South America, and post-communist Europe; O’Donnell and Schmitter, Transitions from authoritarian rule: Tentative conclusions about uncertain democracies. 43 Ziblatt, “How did Europe democratize?,” 326. 44 See Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?”; Bermeo, “On democratic backsliding.” 45 Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?” 46 It is also, notably, the mirror of democratization - or the substantial improvement of electoral democracy over a sustained period of time. See Wilson, et al. “Successful and Failed Episodes of Democratization: Conceptualization, Identification, and Description.” 47 Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?”; Wilson et al., “Successful and Failed Episodes of Democratization: Conceptualization, Identification, and Description.”
10
Figure 1. Typology of autocratization, based on Lührmann and Lindberg
Our typology of autocratization episodes is summarized in Figure 1. Compared to
Lührmann and Lindberg, we share their view that autocratization episodes come in two forms
– here labelled democratic and autocratic regression. We use the more neutral term
“regression” for terminological consistency and avoid terms like “autocratic consolidation”
because substantial moves toward harsher authoritarianism in autocracies could also signal
regime instability. Within an episode of autocratization, countries may move between these
subtypes uninterrupted when an initial period of democratic regression results in a democratic
breakdown followed by a subsequent period of autocratic regression.
We focus our attention exclusively on democratic regression as a subtype of
autocratization.48 Once in motion, episodes of democratic regression can have two possible
outcomes: democratic breakdown or averted democratic breakdown. Therefore, Figure 1
does not include democratic breakdown because it represents one of several possible
outcomes.49 Democratic breakdown occurs when a democracy experiences a genuine
transition into autocracy (as defined below). This possibility is indicated by the arrow for
democratic regression crossing over into authoritarianism in Figure 1. When democratic
breakdown is averted, cases remain on the right side of the same arrow within the democratic
regime spectrum, similar Linz’s “re-equilibriation”.50 What happens after a democratic
48 Similar to Tomini and Wagemann, “Varieties of contemporary democratic breakdown and regression: A comparative analysis.” 49 See Figure 1 in Lührmann and Lindberg’s “A third wave of autocratization is here: what is new about it?”. 50 Linz, The breakdown of democratic regimes.
11
breakdown, which may entail a period of autocratic regression, lies outside the scope of this
study.
Operationalizing episodes of autocratization
We operationalize episodes of autocratization using V-Dem’s Electoral Democracy Index
(EDI). The EDI captures the degree to which a country observes Dahl’s institutional
guarantees of polyarchy.51 It is based on over forty unique indicators aggregated using a state-
of-the-art Bayesian IRT model.52 We consider substantial and sustained declines to begin with
an annual EDI drop of at least 0.01, followed by an overall decline of at least 0.10 throughout
the episode. Autocratization is considered ongoing so long as (i) annual EDI declines
continue for at least one out of every five consecutive years, (ii) the EDI does not increase
by 0.03 or greater in a given year, and (iii) the EDI does not gradually increase by 0.10 over a
five-year period. The end date of all episodes is the year the case experienced an annual
decline of at least 0.01 after episode onset and prior to experiencing one of these three
conditions for termination.
We employ these coding rules to construct the Episodes of Regime Transformation
(ERT) dataset, which identifies 293 episodes of autocratization occurring in 128 political units
from 1900 to 2019.53 The complete ERT dataset also includes 427 episodes of
democratization, as the mirror of autocratization, occurring in 166 political units during the
same time period.
The ERT builds on Lührmann and Lindberg’s operationalization of autocratization
episodes with three modifications: First, we increase the time tolerance from four to five years
to allow more countries to hold an election during the episode interval.54 Second, we increase
the annual upturn threshold from 0.02 to 0.03 because face-validity checks show that the
lower threshold55 can artificially terminate longer episodes (e.g. India 2002–2019, Turkey
2007–2019, and Venezuela 1999–2019) due to a single year increase followed by continued
51. Teorell et al., “Measuring polyarchy across the globe, 1900–2017.”; Dahl, Polyarchy: Participation and opposition. 52 Pemstein et al., “The V-Dem Measurement Model: Latent Variable Analysis for Cross-National and Cross-Temporal Expert-Coded Data”. 53 We provide an R-package that replicates all episodes based on the most recent V-Dem dataset. The package allows for further robustness tests and has flexible parameter settings to redefine the episode data. Maerz et al., “Vdemdata – and R package to load, explore and work with the most recent V-Dem (Varieties of Democracy) dataset”; Edgell et al. “Episodes of Regime Transformation (ERT) dataset codebook, v1.0”. 54 Most countries hold legislative elections every four or five years, with an average term of 4.7 years (cf. Inter-Parliamentary Union, Parliaments at a glance: Term). 55 Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?”
12
sustained declines. Finally, we introduce an additional termination criterion based on gradual
upward changes of 0.10 to reduce overlap with democratization episodes.56
Driven by our motivation to explain democratic resilience as a two-stage process, we
are the first to systematically differentiate democratic and autocratic regression within
episodes of autocratization. To do so, the ERT dataset uses the Regimes of the World (RoW)
classification.57 Because noise can cause some countries to jitter around the RoW cutoffs, we
require that cases reaching the threshold for democracy hold a founding election for the
legislature, executive, or constituent assembly before being considered a democracy.58 We
differentiate the democratic and autocratic regression subtypes based on this modified regime
classification. Of the 293 episodes of autocratization, 96 (38%) include a period of democratic
regression. The remaining 157 autocratization episodes are classified as autocratic regression
and fall outside the scope of this study.
We are also the first to operationalize the outcomes of democratic regression using
the episodes approach. Democratic breakdown occurs when a democratic regime through an
episode of autocratization becomes reclassified as an autocracy. Similar to above, noise in the
data may cause some cases to move into and out of electoral autocracy due to cut points in
the RoW measure.59 We impose three criteria to signal a genuine democratic breakdown: (i)
the country becomes a closed autocracy; (ii) it holds an election while being coded as an
electoral autocracy indicating that the country is now a de facto electoral autocracy; or (iii) it
becomes an electoral autocracy and stays that way for at least 5 years. Episodes of democratic
regression where the regime avoided becoming autocratic before the end of the episode are
coded as averted democratic breakdown for their outcome. Based on these criteria, 65 (68%)
episodes of democratic regression have resulted in democratic breakdown since 1900, and
only 19 (20%) averted democratic breakdown. This suggests that few democracies survive
once an episode of democratic regression is underway.60
An alternative autocratization operationalization is found in the annual Democracy
Report and “State of the World” articles from the V-Dem Institute.61 It measures
autocratization in a simplified manner as substantial if there is a drop of at least 0.05 on the
56 Wilson et al., “Successful and Failed Episodes of Democratization: Conceptualization, Identification, and Description.” 57 Lührmann, Tannenberg, and Lindberg, “Regimes of the World (RoW): Opening New Avenues for the Comparative Study of Political Regimes.” 58 Also see Wilson et al., “Successful and Failed Episodes of Democratization: Conceptualization, Identification, and Description.” 59 Kasuya and Mori, “Better Regime Cutoffs for Continuous Democracy Measures.” 60 In the current version, 12 episodes of democratic regression are censored because their outcome is yet undetermined. 61 E.g. Maerz et al., “State of the world 2019: Autocratization Surges - Resistance Grows.”
13
Liberal Democracy Index (LDI) between the start- and end-year of a ten year period.62 While
this approach offers a broad and easy-to-grasp operationalization suitable for the general
audience these publications are designated for, the present episodes-based approach with
(admittedly somewhat complex) coding decisions driven by theories from the literature on
regime change, provides a much more rigorous foundation for research. It also allows for
analysis of the two possible forms of democratic resilience - i.e. avoiding autocratization
episodes altogether and averting democratic breakdown once in an episode.
Descriptive analysis
Figure 2 plots global trends for the two subtypes of autocratization episodes from 1900 to
2019. In the Appendix, Figure 6 compares trends in episodes of autocratization and
democratization from 1900 to 2019. Together, these two figures provide an update on the
progression of the “third wave of autocratization”,63 including additional detail on the types
of episodes within each wave. Generally, the results suggest that autocratization episodes are
relatively uncommon, but their prevalence is increasing over time. The “third wave of
autocratization” is particularly intense, with a record 25% of polities experiencing an episode
of autocratization in 2017 (the sum of both types in Figure 2). Since then, the share of
countries in autocratization episodes has declined to about 21%. This contrasts with the
previous second wave of autocratization, which affected about 15% of the world’s countries
at its peak in 1965. Figure 2 also suggests that the nature of autocratization is changing and
threatens democracies more than in previous waves. Of the 96 episodes of democratic
regression, 59 (61%) began after 1992. Apart from a few cases in the 1930s, autocratization
in democracies is overwhelmingly a post-Cold War phenomenon.
62 The LDI aggregates V-Dem’s Electoral Democracy Index (EDI) and Liberal Component Index (LCI). 63 Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?”
14
Figure 2. Episodes of democratic and autocratic regression, 1900--2019
The increasing share of autocratization episodes in democracies may be partially
explained by a democratic world. At the start of the third wave of autocratization in 1993,
41% of countries were democracies, compared to just 19% at the start of the second wave
and 15% at the start of the first wave of autocratization. If autocratization is randomly
“assigned” to countries, the third wave should affect a greater number of democracies than
previous waves. However, existing theories suggest that autocratization episodes are not
random processes but spurred by leaders, party elites, and various groups in societies.64
Therefore, the high incidence of democratic regression in the current period may mark a new
and more worrying trajectory for democracies than Lührmann and Lindberg suspected.
While democratic regression has declined recently, this could be driven by a high
fatality rate for democracy. So far, the third wave of autocratization has led to the breakdown
64 Linz, The breakdown of democratic regimes.
15
Figure 3. Trajectories of democratic regression episodes that ended with democratic
breakdown. Red dots mark the start year of an episode and the red crosses mark
the end year. Plots include the pre- and post-episode year. Number of episode by
country in brackets.
of an unprecedented 36 democratic regimes. As a result, 717 million people have lost access
to democratic institutions and freedoms.65 Some of these countries (28) have continued to
regress as autocracies, contributing to an increase in autocratic regression.66
Figure 3 and Figure 4 plot the trajectories of each democratic regression episode in
the ERT dataset differentiated by outcome. The two figures generate three insights. First,
they reinforce the importance of taking an episodes approach to understanding democratic
resilience. The lines for each episode show wide variation in the quality of democracy at the
onset of democratic regression, the degree to which democracy declines throughout the
episode, and the duration of the autocratization process. Taking democratic
65 We sum the population of all countries experiencing autocratization according to our definition after 1992. 66 Three breakdowns in 2019 – Bolivia, Philippines, and Benin – are also likely to experience subsequent autocratic regression in the coming years.
16
Figure 4. Trajectories of democratic regression episodes that ended without democratic
breakdown. Red dots mark the start year of an episode and the red crosses mark
the end year. Plots include the pre- and post-episode year. Number of episode by
country in brackets.
survival, breakdown, or annual changes at a given point in time would obscure this variation
and the potentially vital information from the patterns that can help explain democratic
resilience. Second, they further illustrate the high prevalence of democratic regression in the
most recent wave of autocratization regardless of outcome (see Appendix, Figure 6). Third,
while the third wave poses a high risk of democratic breakdown, Figure 4 reveals that averted
breakdown is also much more common now than in past waves of autocratization. Prior to
1992, only five democracies undergoing an episode of autocratization managed to avert
democratic breakdown – Australia (1917), Finland (1940), France (1965), India (1976), and
Trinidad and Tobago (1972). Since then, fourteen others have followed suit.
Explaining democratic resilience
Data and empirical approach
Following our conceptualization of democratic resilience as a two-stage phenomenon, we use
17
a Heckman selection model to estimate determinants of democratic regression and
breakdown.67 The first “selection” stage estimates the probability that a given democratic
country-year falls within an autocratization episode. Thus, the outcome variable equals one
while being in an autocratization episode and zero otherwise, using a sample of democratic
country-years within the ERT dataset (3,759 observations). The second “outcome” stage
estimates the probability that a given year of democratic regression results in democratic
breakdown. This model is conditional on the case being within an episode of democratic
regression, accounting for selection bias estimated in the first stage. Thus, the outcome
variable is coded as one for democratic breakdown and zero otherwise, with a sample of all
country-years coded as one in the first stage (330 observations).68 To reduce concerns of
simultaneity bias, that could arise if aspiring autocrats dismantle institutional checks and
balances, we lag all variables (except for coups) by one year.
We focus on a number of economic and political variables that emerge from the
literature as prime suspects for explaining democratic resilience. First, to capture the key
mechanism in the “perils of presidentialism”, we include two de facto measures of executive
constraints provided by the V-Dem dataset: the judicial constraints on the executive index
(v2x_jucon) and the legislative constraints on the executive index (v2xlg_legcon).69 The
former measures judicial independence and whether the executive respects court rulings and
the constitution. The latter indicates the degree to which the legislature and government
agencies exercise oversight of the executive.70 Second, we include measures of inflation-
adjusted GDP per capita and economic growth from the Maddison project71 to capture level
of economic development and economic performance, respectively. Third, to address spatial
clustering of regimes and potential neighbourhood effects found in the literature, we include
the average scores of V-Dem’s EDI for all other countries in the region using the tenfold
geo-political classification scheme in V-Dem (e_regionpol).72 Finally, studies show that
democracies in countries with a longer history of democratic rule are more likely to survive.73
67 Toomet and Henningsen, “Sample Selection Models in R: Package sampleSelection.” 68 For details on our episode coding criteria, refer to the ERT codebook. 69 Linz, “The perils of presidentialism.” 70 While both indices are part of the Liberal Component Index, there is no overlap with the EDI that is used to determine the start and end of autocratization episodes. For the full list of all variables included in the two indices, see Coppedge et al. (V-Dem Codebook v10, 357). 71 Bold et al., “Rebasing ‘Maddison’: new income comparisons and the shape of long-run economic development.” We use a five-year moving average of GDP growth to make sure our results are not driven by short-term fluctuations. 72 E.g. Brinks and Coppedge, “Diffusion is no illusion: Neighbor emulation in the third wave of democracy,” 73 E.g. Svolik, “Authoritarian reversals and democratic consolidation.”
18
To capture past democratic experience, we draw on a recently developed measure of
democratic stock that captures the accumulation of democratic experience over time.74
In addition to this common set of predictors and in line with assumptions of the
Heckman selection model, we include some additional predictors for “selection into”
democratic regression in the first stage. First, we count the cumulative number of previous
autocratization episodes. A large number of previous episodes should be indicative of a high
general vulnerability to democratic regression. Second, as evidenced from the descriptive
analysis above, regime transformations unfold in global waves. Therefore, we account for the
percentage of countries with ongoing democratization and autocratization episodes for each
year. Third, we include region dummies in the first stage to control for unobserved time-
invariant factors at the regional level that affect autocratization.75 These additional selection
variables are only included in the first stage of the model to satisfy the exclusion restriction.
We argue that these variables determine whether a country is more likely to autocratize, for
instance, if it is hit by a global wave of autocratization, but that they are substantively
unrelated to the outcome of the episode.
We also control for a series of other correlates of democratic resilience. Because
military coups are one of the main threats to democracy,76 we control for the occurrence of
one or more military coups in a country (binary indicator) by combining information from
two coup datasets included in V-Dem (e_pt_coup and e_coups).77 We also include population
size from the Maddison project78 as it might affect a polity’s susceptibility to conflict and
democratic regression. In the second stage, we control for the duration of the episode by
including the number of years since episode onset and its square term, as shorter or longer
episodes may be more prone to breakdown.79 We include decade dummies in both stages of
the model to account for global shocks such as the two World Wars simultaneously affecting
a large number of countries. Summary statistics for all variables in the analysis are displayed
in Table 2 (full sample) and Table 3 (episodes sample) in the Appendix.
74 Edgell et al., “Democratic Legacies: Using Democratic Stock to Assess Norms, Growth, and Regime Trajectories.” 75 Due to the low number of observations in the second stage, we cannot include region dummies in the second stage. 76 Marinov and Goemans, “Coups and democracy.” 77 Powell and Thyne, “Global instances of coups from 1950 to 2010: A new dataset”; Przeworski et al., Political Institutions and Political Events (PIPE) Data Set. 78 Bold et al., “Rebasing ‘Maddison’: new income comparisons and the shape of long-run economic development.” 79 The second stage is similar to a regular duration model, which is why we add a variable for the duration of the ongoing episode as well as its squared term.
19
Table 1. Main results: Heckman-style selection model
Dependent variable in selection equation: democratic regression.
Dependent variable in outcome equation: democratic breakdown.
Standard errors clustered at the country-level. Significance levels ***
p < 0.001, ** p < 0.01, * p < 0.05.
20
Results
The results are summarized in Table 1. The left column shows the probit regression
coefficients for the first stage of the model (selecting into a democratic regression episode).
In line with our expectations regarding institutional checks and balances, the results
demonstrate that greater legislative and judicial constraints on the executive are associated
with a decreased probability of experiencing democratic regression. The results also support
the hypothesis that a higher level of economic development is linked to significant reductions
in the likelihood of experiencing an episode of democratic regression. Economic
performance, however, does not show a statistically significant effect. While regional levels
of democracy and democratic stock display the expected negative coefficient, their effect is
not statistically significant in the first stage of the model. We can thus corroborate the
relevance of political and economic factors in explaining resilience to democratic regression.
Among the selection variables included in the first stage, we see that previous autocratization
episodes make democracies more prone to experience democratic regression and a higher
percentage of democratizing countries globally decreases the likelihood of democratic
regression.
The right column of Table 1 shows the results for the outcome stage of the model.
Most variables show results similar to the first stage. Judicial constraints on the executive
significantly reduce the likelihood of breakdown, while the effect for legislative constraints
on the executive is not significant in the second stage. The importance of different institutions
varies over the course of democratic regression. Judicial institutions act as the last bulwark
against democratic breakdown. Our results further show that economic development and
growth do not offer much explanation for why some democracies break down and others do
not. However, higher levels of democratic stock and quality of democracy in neighbouring
countries significantly reduce the chance of regime breakdown.
We simulate predicted probabilities for ongoing democratic regression episodes (first
stage) and democratic breakdown (second stage) based on our model estimates and plot them
over the range of the key independent variables in Figure 5. The plots on the left show how
the probability of experiencing democratic regression varies with increasing legislative and
judicial constraints on the executive and GDP per capita. Country-years where the de facto
constraints on the executive are greater, are much less
21
Figure 5. Predicted probabilities of being in an episode of democratic regression (left panel)
and democratic breakdown (right panel) over the range of selected explanatory
variables. Estimates and 95% confidence intervals are based on simulations from
the model parameters.
22
likely to be followed by an episode of democratic regression. At low levels of executive
constraints, there is a 10% chance that democratic regression occurs, while the chances are
almost zero with tight constraints.
The relationship between economic development and episode onset is even more
pronounced. Democracies at the lower end of the income distribution like Niger or Liberia
face a considerable risk of experiencing democratic regression whereas this almost never
occurs in very rich countries. A comparison of the plots on the left and right underlines the
usefulness of separating ongoing episodes and episode outcomes. While legislative and
judicial constraints make onset less likely, only judicial constraints are related to a lower
likelihood of democratic breakdown conditioned on that a democratic regression began.
We report a series of robustness tests in the Appendix. First, we run separate models
for episode onset and outcome. We argue that democratic regression is a phenomenon
potentially unfolding over several stages and years. However, the Heckman model does not
properly model episode onset; it merely distinguishes countries with ongoing episodes from
those that do not experience democratic regression. Therefore, we run an event history model
(Model 2 in Table 4) in which we exclude all ongoing episode-years and add polynomials for
time since last democratic regression (or initial democratic transition). The results for episode
onset are highly comparable to the first stage of our selection model and corroborate that our
main independent variables are valid predictors of episode onset.80
Second, we rerun our main model including all components of the executive
constraints indices as separate predictors to test if certain sub-components are particularly
influential (Model 4 and Model 5 in Table 5). Only indicators for executive respects
constitution and legislature questions officials are significantly associated with a lower
likelihood of experiencing autocratization. The results for all other variables remain
unchanged.
Third, our operationalization of episodes depends on several, arguably idiosyncratic
choices regarding how to identify start and end dates of episodes. While we have chosen
theoretically motivated default parameters, we run additional models where we modify
these.81 For Model 7 in Table 6, we reduce the threshold to avert democratic breakdown in
one year from 0.03 to 0.02 in line with Lührmann and Lindberg. For Model 8, we lower the
threshold for total decline to 0.05. The results corroborate our initial results.
80 Model 3 (Table 4) is a probit model of the second stage of our selection model with similar results even when ignoring the selection process. 81 Our accompanying R package makes it easy for the user to operationalize episodes according to his or her own criteria.
23
A more general challenge is the small number of episodes. Thus, including a large
number of explanatory variables can be problematic, as can the exclusion or inclusion of
influential episode cases. However, our main findings are robust to different modelling
choices and operationalization of autocratization episodes. Executive constraints are
consistently associated with a lower likelihood of democratic regression and democratic
breakdown. While economic development can prevent democratic regression, it cannot
explain why some countries break down while others do not. By contrast, countries with
more democratic stock and higher levels of democracy in their neighbourhood are less likely
to see democratic breakdown.
Conclusion
The world is currently experiencing a third wave of autocratization characterized by the
gradual erosion of democratic norms and executive aggrandizement.82 Yet, existing
quantitative studies of democratic resilience typically operationalize democratic breakdown
as events. This disregards conceptual and empirical differences between those democracies
that never experience autocratization and those that having begun autocratizing somehow
manage to avert breakdown. This naturally leads to questions about selection bias, especially
if factors influencing the experience of an autocratization episode are correlated with the
outcome of the episode itself. This means that our existing theories remain incomplete until
we account for the two-stage nature of democratic resilience.
Thus, this article expands our understanding of democratic resilience by taking an
episodes approach to studying autocratization. Such an approach enables us to fully account
for the two forms of democratic resilience by conceptualizing autocratization as a process
with defined beginning, end, and outcome. Thus, democratic resilience can be analysed in
two stages – either by avoiding democratic regression altogether or, once it has started, by
avoiding a full breakdown. To do so, we develop the ERT dataset, the most comprehensive
identification of episodes of autocratization from 1900-2019, along with an R-package
enabling many varied analyses of the phenomena. This includes a refined empirical
delineation of democratic regression episodes allowing for a distinction between cases in
which democracy broke down from those in which breakdown was averted.
82 Lührmann and Lindberg, “A third wave of autocratization is here: what is new about it?”; Bermeo, “On democratic backsliding.”
24
Through our descriptive analysis, we demonstrate that for democracies,
autocratization episodes are overwhelmingly a post-Cold War phenomenon. Of the 96
episodes of democratic regression in the ERT dataset, 59 (61%) began after 1992. Thus far,
the third wave of autocratization has led to the breakdown of 36 democratic regimes, an
average of 1.33 per year.83 Never before has the world witnessed such a high fatality rate for
democracy. This suggests that the current third wave of autocratization has an extraordinary
effect on democracies.
Results from the explanatory models provide insights into how democracies can be
resilient to this wave of democratic regression. In particular, how democracies prevail is partly
a function of institutions, economics, and the political environment. However, these factors
have different effects depending on the stage of the process. Both, legislative and judicial
constraints, for example, are associated with a decreasing likelihood of experiencing
democratic regression. Yet, once democratic regression is ongoing only judicial constraints
matter for averting democratic breakdown. This suggests that legislatures may be more
affected during the early stages of democratic regression, to the point of being obsolete as
bulwarks safeguarding democracy by the time breakdown occurs. By contrast, judicial
institutions may play an important role as democracy’s last line of defence. The findings also
suggest that economic development is associated with a reduced likelihood of undergoing
democratic regression but has little influence on the outcome once an episode has begun.
Instead, to avert breakdown once democratic regression is ongoing, the results of this study
suggest that having democratic neighbours and long previous democratic experiences are key.
This study further underscores the need for more holistic research on the changing
role of structural factors in different stages of regime transformation. For practitioners in
democracy promotion and pro-democracy activists, the spectre of autocratization requires
different responses depending on whether the process has already begun. Only then can
democracies prevail.
83 As compared to 0.73 per year in the first wave and 0.21 per year in the second wave of autocratization.
25
References
Alemán, José, and David D. Yang. “A duration analysis of democratic transitions and
authoritarian backslides.” Comparative Political Studies 44, no. 9 (2011): 1123–1151.
Bader, Julia, Jörn Grävingholt, and Antje Kästner. “Would autocracies promote autocracy? A
political economy perspective on regime-type export in regional neighbourhoods.”
Contemporary Politics 16, no. 1 (2010): 81–100.
Bermeo, Nancy. “On democratic backsliding.” Journal of Democracy 27, no. 1 (2016): 5–19.
Bermeo, Nancy G. Ordinary people in extraordinary times: The citizenry and the breakdown of democracy.
Princeton University Press, 2003.
Bernhard, Michael, Timothy Nordstrom, and Christopher Reenock. “Economic
performance, institutional intermediation, and democratic survival.” Journal of Politics 63,
no. 3 (2001): 775–803.
Boix, Carles, and Susan C Stokes. “Endogenous democratization.” World Politics 55, no. 4
(2003): 517–549.
Bold, Jutta, Robert Inklaar, Herman de Jong, and Jan Luiten van Zanden. “Rebasing
‘Maddison’: new income comparisons and the shape of long-run economic
development.” Maddison Project Working Paper, no. 10 (2018).
Brinks, Daniel, and Michael Coppedge. “Diffusion is no illusion: Neighbor emulation in the
third wave of democracy.” Comparative Political Studies 39, no. 4 (2006): 463–489.
Burkhart, Ross E., and Michael S. Lewis-Beck. “Comparative Democracy: The Economic
Development Thesis.” American Political Science Review 88, no. 4 (1994): 903–910.
Burnell, Peter, and Peter Calvert. “The resilience of democracy: An introduction.”
Democratization 6, no. 1 (1999): 1–32.
Cheibub, José Antonio. Presidentialism, parliamentarism, and democracy. New York: Cambridge
University Press, 2007.
Cheibub, José Antonio, Adam Przeworski, and Sebastian M Saiegh. “Government coalitions
and legislative success under presidentialism and parliamentarism.” British Journal of
Political Science 34, no. 4 (2004): 565–587.
Coppedge, Michael. “Eroding regimes: What, where, and when?” V-Dem Working Paper 57
(2017).
Coppedge, Michael, John Gerring, Carl Henrik Knutsen, Staffan I. Lindberg, Jan Teorell,
David Altman, Michael Bernhard, et al. V-Dem Codebook v10. Varieties of Democracy
(V-Dem) Project, 2020.
Coppedge, Michael, John Gerring, Carl Henrik Knutsen, Staffan I. Lindberg, Jan Teorell,
26
David Altman, Michael Bernhard, et al. “V-Dem Dataset V10,” 2020.
Dahl, Robert A. Polyarchy: Participation and opposition. New Haven: Yale University Press, 1971.
Edgell, Amanda B., Vanessa A. Boese, Seraphine F. Maerz, Patrik Lindenfors, and Staffan I.
Lindberg. “Establishing Pathways to Democracy Using Domination Analysis.” V-Dem
Working Paper Series, No. 95, 2019.
Edgell, Amanda B., Matthew C. Wilson, Vanessa A. Boese, and Sandra Grahn. “Democratic
Legacies: Using Democratic Stock to Assess Norms, Growth, and Regime Trajectories.”
V-Dem Users Working Paper Series, No. 100, 2020.
Edgell, Amanda B., Seraphine F. Maerz, Laura Maxwell, Richard Morgan, Juraj Medzihorsky,
Matthew C. Wilson, Vanessa Boese, Sebastian Hellmeier, Jean Lachapelle, Patrik
Lindenfors, Anna Lührmann, and Staffan I. Lindberg. (2020). “Episodes of Regime
Transformation Dataset, v1.0.” Varieties of Democracy (V-Dem) Institute. 2020.
https://github.com/vdeminstitute/vdemdata
Epstein, David L, Robert Bates, Jack Goldstone, Ida Kristensen, and Sharyn O’Halloran.
“Democratic transitions.” American Journal of Political Science 50, no. 3 (2006): 551–569.
Erdmann, Gero. “Transition from Democracy. Loss of Quality, Hybridisation and
Breakdown of Democracy.” Comparative Governance and Politics, no. 1 (2011): 21–58.
Feng, Yi. “Democracy, political stability and economic growth.” British Journal of Political Science
27, no. 3 (1997): 391–418.
Gasiorowski, Mark J. “Economic Crisis and Political Regime Change: An Event History
Analysis.” American Political Science Review 89, no. 4 (1995): 882–897.
Gates, Scott, Håvard Hegre, Mark P. Jones, and Håvard Strand. “Institutional Inconsistency
and Political Instability: Polity Duration, 1800–2000.” American Journal of Political Science
50, no. 4 (2006): 893–908.
Gerring, John. “Mere description.” British Journal of Political Science 42, no. 4 (2012): 721– 746.
Ginsburg, Tom, and Aziz Z Huq. How to save a constitutional democracy. Chicago: University of
Chicago Press, 2018.
Gleditsch, Kristian Skrede. All international politics is local: The diffusion of conflict, integration, and
democratization. University of Michigan Press, 2009.
Gleditsch, Kristian Skrede, and Michael D Ward. “Diffusion and the international context of
democratization.” International Organization 60, no. 4 (2006): 911–933.
Grosfeld, Irena, and Ekaterina Zhuravskaya. “Cultural vs. economic legacies of empires:
Evidence from the partition of Poland.” Journal of Comparative Economics 43, no. 1 (2015):
55–75.
27
Guiso, Luigi, Paola Sapienza, and Luigi Zingales. “Long-term persistence.” Journal of the
European Economic Association 14, no. 6 (2016): 1401–1436.
Hiroi, Taeko, and Sawa Omori. “Perils of parliamentarism? Political systems and the stability
of democracy revisited.” Democratization 16, no. 3 (2009): 485–507.
Houle, Christian, and Mark A. Kayser. “The Two-step Model of Clustered Democratization.”
Journal of Conflict Resolution 63, no. 10 (2019): 2421–2437.
Hollyer, James R, Peter B. Rosendorff and James R. Vreeland, “Transparency, Protest and
Democratic Stability.” British Journal of Political Science 49, no. 4 (2019): 1251-1277.
Inter-Parliamentary Union. Parliaments at a glance: Term, 2020. http://archive.ipu.org/ parline-
e/TermofParliament.asp?REGION=All&LANG=ENG&typesearch=1.
Kasuya, Yuko, and Kota Mori. “Better Regime Cutoffs for Continuous Democracy
Measures.” V-Dem Users Working Paper Series, No. 25, 2019.
Kaufman, Robert R, and Stephan Haggard. “Democratic decline in the United States: What
can we learn from middle-income backsliding?” Perspectives on Politics 17, no. 2 (2019):
417–432.
Laebens, Melis, and Anna Lührmann. “What stops democratic erosion? The role of
institutions of accountability.” Paper prepared for the Berlin Democracy Conference, Berlin, 11-
12 November 2019, 2019.
Lijphart, Arend. “Constitutional design for divided societies.” Journal of Democracy 15, no. 2
(2004): 96–109.
Lijphart, Arend. Democracy in plural societies: A comparative exploration. New Haven, CT: Yale
University Press, 1977.
Lindberg, Staffan I. Democracy and elections in Africa. Baltimore, MD: Johns Hopkins University
Press, 2006.
Linz, Juan J. “The perils of presidentialism.” Journal of Democracy 1, no. 1 (1990): 51–69.
Linz, Juan J, and Alfred Stepan. Problems of democratic transition and consolidation: Southern Europe,
South America, and post-communist Europe. Baltimore, MD: Johns Hopkins University Press,
1996.
Linz, Juan J, and Arturo Valenzuela. The failure of presidential democracy. Vol. 1. Baltimore, MD:
Johns Hopkins University Press, 1994.
Linz, Juan J. The breakdown of democratic regimes. Johns Hopkins University Press, 1978.
Lipset, Seymour Martin. “Some social requisites of democracy: Economic development and
political legitimacy.” American Political Science Review 53, no. 1 (1959): 69–105.
Lührmann, Anna, and Staffan I. Lindberg. “A third wave of autocratization is here: what is
28
new about it?” Democratization, 2019, 1095–1113.
Lührmann, Anna, Marcus Tannenberg, and Staffan I Lindberg. “Regimes of the World
(RoW): Opening New Avenues for the Comparative Study of Political Regimes.” Politics
& Governance 6, no. 1 (2018).
Maerz, Seraphine F., Anna Lührmann, Sebastian Hellmeier, Sandra Grahn, and Staffan I. Lindberg.
“State of the world 2019: Autocratization Surges - Resistance Grows.” Forthcoming in
Democratization, 2020. Online First. https://doi.org/10.1080/13510347.2020.1758670
Maerz, Seraphine F., Amanda B. Edgell, Joshua Krusell, Laura Maxwekk, Sebastian Hellmeier,
and Nina Ilchenko. “Vdemdata – an R package to load, explore and work with the most
recent V-Dem (Varieties of Democracy) dataset.” Varieties of Democracy (V-Dem) Project,
2020. https://github.com/vdeminstitute/vdemdata.
Mainwaring, Scott. “Presidentialism, multipartism, and democracy: The difficult
combination.” Comparative Political Studies 26, no. 2 (1993): 198–228.
Mainwaring, Scott, and Anıbal Pérez-Linñán. Democracies and dictatorships in Latin America:
emergence, survival, and fall. Cambridge University Press, 2013.
Marinov, Nikolay, and Hein Goemans. “Coups and democracy.” British Journal of Political
Science 44, no. 4 (2014): 799–825.
Mechkova, Valeriya, Anna Lührmann, and Staffan I Lindberg. “How much democratic
backsliding?” Journal of Democracy 28, no. 4 (2017): 162–169.
Morlino, Leonardo, and Mario Quaranta. “What is the impact of the economic crisis on
democracy? Evidence from Europe.” International Political Science Review 37, no. 5 (2016):
618–633.
O‘Donnell, Guillermo, and Philippe C. Schmitter. Transitions from authoritarian rule: Tentative
conclusions about uncertain democracies. Baltimore, MD: Johns Hopkins University Press,
1986.
Olson, Mancur. “Dictatorship, Democracy, and Development.” American Political Science Review
87, no. 3 (1993): 567–576.
Pemstein, Daniel, Kyle L. Marquardt, Eitan Tzelgov, Yi-ting Wang, Juraj Medzihorsky, Joshua
Krusell, Farhad Miri, and Johannes von Römer. “The V-Dem Measurement Model:
Latent Variable Analysis for Cross-National and Cross-Temporal Expert-Coded Data.”
V-Dem Working Paper No. 21. 4th edition. University of Gothenburg: Varieties of Democracy
Institute, 2019.
Pérez-Liñán, Anıbal, and Scott Mainwaring. “Regime legacies and levels of democracy:
evidence from Latin America.” Comparative Politics 45, no. 4 (2013): 379–397.
Pérez-Liñán, Anıbal, Nicolás Schmidt, and Daniela Vairo. “Presidential hegemony and demo-
29
cratic backsliding in Latin America, 1925–2016.” Democratization 26, no. 4 (2019): 606–
625.
Persson, Torsten, and Guido Tabellini. “Democratic capital: the nexus of political and
economic change.” American Economic Journal: Macroeconomics 1, no. 2 (2009): 88–126.
Powell, Jonathan M, and Clayton L Thyne. “Global instances of coups from 1950 to 2010: A
new dataset.” Journal of Peace Research 48, no. 2 (2011): 249–259.
Przeworski, A., S. Newman, S. K. Park, D. Queralt, G. Rivero, and K. J. Shin. Political
Institutions and Political Events (PIPE) Data Set. Department of Politics, New York
University, New York.
Przeworski, Adam, R Michael Alvarez, Michael E Alvarez, Jose Antonio Cheibub, Fernando
Limongi. Democracy and development: Political institutions and well-being in the world, 1950-1990.
Vol. 3. Cambridge University Press, 2000.
Przeworski, Adam, and Fernando Limongi. “Modernization: Theories and facts.” World politics
49, no. 2 (1997): 155–183.
Putnam, Robert. Making Democracy Work: Civic Traditions in Modern Italy. 1–280. Princeton:
Princeton University Press, 1993.
Reenock, Christopher, Jeffrey K Staton, and Marius Radean. “Legal institutions and
democratic survival.” The Journal of Politics 75, no. 2 (2013): 491–505.
Riggs, Fred W. “Presidentialism versus parliamentarism: Implications for representativeness
and legitimacy.” International Political Science Review 18, no. 3 (1997): 253–278.
Risse, Thomas, and Nelli Babayan. “Democracy promotion and the challenges of illiberal
regional powers: introduction to the special issue.” Democratization 22, no. 3 (2015): 381–
399.
Stepan, Alfred, and Cindy Skach. “Constitutional frameworks and democratic consolidation:
Parliamentarianism versus presidentialism.” World Politics 46, no. 1 (1993): 1–22.
Svolik, Milan. “Authoritarian reversals and democratic consolidation.” American Political Science
Review 102, no. 2 (2008): 153–168.
Svolik, Milan W. “Which democracies will last? Coups, incumbent takeovers, and the dynamic
of democratic consolidation.” British Journal of Political Science 45, no. 4 (2015): 715–738.
Tansey, Oisín, Kevin Koehler, and Alexander Schmotz. “Ties to the Rest: Autocratic
Linkages and Regime Survival.” Comparative Political Studies 50, no. 9 (2017): 1221–1254.
Teorell, Jan. Determinants of democratization: Explaining regime change in the world, 1972– 2006.
Cambridge University Press, 2010.
Teorell, Jan, Michael Coppedge, Staffan I. Lindberg, and Svend-Erik Skaaning. “Measuring
30
polyarchy across the globe, 1900–2017.” Studies in Comparative International Development 54,
no. 1 (2019): 71–95.
Tilly, Charles. Contention and democracy in Europe, 1650-2000. New York: Cambridge University
Press, 2004.
Tomini, Luca, and Claudius Wagemann. “Varieties of contemporary democratic breakdown
and regression: A comparative analysis.” European Journal of Political Research 57, no. 3
(2018): 687–716.
Toomet, Ott, and Arne Henningsen. “Sample Selection Models in R: Package sample
Selection.” Journal of Statistical Software 27, no. 7 (2008).
Wilson, Matthew C., Richard Morgan, Juraj Medzihorsky, Laura Maxwell, Seraphine F. Maerz,
Anna Lührmann, Patrik Lindenfors, Amanda B. Edgell, Vanessa Boese, and Staffan I.
Lindberg. “Successful and Failed Episodes of Democratization: Conceptualization,
Identification, and Description.” V-Dem Working Papers, no. 79 (2020).
Wilson, Matthew Charles. “A Closer Look at the Limits of Consociationalism.” Comparative
Political Studies 53, no. 5 (2020): 700–733.
Ziblatt, Daniel. “How did Europe democratize?” World Politics 58, no. 2 (2006): 311–338.
first wave of
democratization
second wave of
democratization
third wave of
democratization
first wave of
autocratization
second wave of
autocratization
third wave of
autocratization
0
10
20
30
40
1900 1920 1940 1960 1980 2000 2020year
cou
ntri
es (
%)
autocratization democratization
